Effects of a high-quality protein diet on glycemic control in patients with prediabetes: a systematic review and network meta-analysis

Aim Prediabetes markedly raises the risk of diabetes and cardiovascular complications. High-quality protein diets benefit glucose and lipid metabolism, but protein sources have not been compared systematically. This network meta-analysis (NMA) compared them across core metabolic outcomes in adults with prediabetes, to inform individualized nutrition. Methods Electronic searches of Cochrane Library, Embase, PubMed, and Web of Science extended to February 10, 2026. We designated FPG (mmol/L) and HbA1c (%) as primary outcomes, with fasting insulin (μIU/mL), HOMA-IR, BMI (kg/m 2 ), total cholesterol (mg/dL), LDL-C (mg/dL), and triglycerides (mg/dL) as secondary outcomes. Bayesian NMA was performed using R 4.4.1, with effects expressed as MD and 95% CrI. Results In total, 18 studies (956 patients) were included. Mixed animal protein (MAP) reduced FPG most effectively (MD -0.30, 95% CrI -0.57 to -0.029). Legume protein (LP) was most effective for HOMA-IR (MD -0.49, 95% CrI -0.55 to -0.43), lamb + chicken protein (LCP) for total cholesterol (MD -7.6, 95% CrI -14 to -1.4), and fish protein (FP) for LDL-C (MD -7.7, 95% CrI -15 to -0.64). No diet significantly affected HbA1c, fasting insulin, BMI, or triglycerides. Conclusions LP was significantly superior to a low-protein diet for HOMA-IR; LCP significantly reduced total cholesterol and FP significantly reduced LDL-C. No other intervention reached significance. These differences may reflect the nutrients accompanying each protein source rather than protein content alone. Given the few primary studies, heterogeneity among some interventions, and short intervention periods, further well-reported trials and implementation research are needed to bridge evidence and practice. Systematic review registration https://www.crd.york.ac.uk/PROSPERO/myprospero , identifier CRD420261325032.